Determine the geographic center

[prravich]

Hi all,

I have customer geographic data that has latitudes and longitudes and I am looking for a way to find the “Center of Gravity” for this data (i.e., one point which has the access to the most demand).  Can you tell me how to approach this ?

Also , If I decide to have more than one center of gravity ( i.e , 2-3 points to get maximum coverage to demands) what would  be the best way?

I am completely new to Spatial analytics so any examples would be appreciated.

Thanks

[MarqueeCrew · Accepted solution]

@prravich

Here is what you "can" do:

1. CREATE POINTS:  Converts lat/lon to spatial object as "centroid"
2. POLY-BUILD: Build a convex Hull as a polygon that includes all of your centroids (optionally, you can group and create many polygons).
3. SPATIAL INFO: Output the Centroid as either (or both) Spatial Object and X & Y fields

All done....

Cheers,

Mark

[prravich]

Thank you I tried and it works. That was so simple

I am still unclear on how to determine the second ask. Assume i have 10 customers randomly scattered, what are the three best places to place my store to get good coverage ? I need to create three polygons and find their center, but not sure how to proceed. Any thoughts?

[MarqueeCrew]

Luckily you can have many solutions marked on a single post.  Unluckily, this second question is for someone with more spatial experience than I have.  I don't think that division by straight lines is the way to go as you don't want to be in the middle of Lake Michigan.  Clustering the customers by count isn't right either.  There is an art to writing that requirement.  I'll tag a couple of members who take up a lot of space (spatial problems).

@NicoleJohnson and @AndyMoncla

[prravich]

Thank you once again. 

[NicoleJohnson]

Do you have some sample spatial data you can provide for, say, 10 (pretend) customers?

I have a feeling the solution will involve the Make Grid tool, but it's hard to test my theory without sample data... essentially the Make Grid tool can split up your polygon from Step 2 in @MarqueeCrew solution (you can define how big/small you want your grid to be, then choose the option to Generate Grids for Each Object), and then by finding the Centroid of each grid, you could use the Find Nearest tool to figure out which grid centers are closest to which potential store locations... but from there it gets a bit tricky with sorting & sampling for the top locations, so it would be helpful to have some sample data to play with... 

Let me know if you can provide, or if the suggestion above gets you pointed in the right direction! 

Cheers,

NJ

[prravich]

Here's a random sample data ,these are the customers I ship to. Based on the geographic location alone , can you help me find 3 places to set up my manufacturing locations to get good coverage?

Thanks

[Dmuller]

I have done the calculation based on the info provided in your file as shown in the attached. Geographically, the centre would be a simple average of the latitudes and longitudes. This will place you about halfway between Iberia and Crocker, Missouri (38.001373,-92.294513)

If you want to take the value into account, you need to determine the weighted average of the longs and lats by using the values to determine the contribution. This will place you in Tonkawa, Oklahoma City (36.605758,-97.237447)